文档报告支持生成小标题功能

zhaoqingang

2024-12-31 6b4093952e555e1eb2713bd85133a5f697cda1e0

文档报告支持生成小标题功能

 app/api/auth.py

@@ -30,14 +30,14 @@
    if db_user:
        return Response(code=200, msg="Username already registered")

    bisheng_service = BishengService(settings.sgb_base_url)
    # bisheng_service = BishengService(settings.sgb_base_url)
    ragflow_service = RagflowService(settings.fwr_base_url)

    # 注册到毕昇
    try:
        bisheng_info = await bisheng_service.register(user.username, user.password)
    except Exception as e:
        return Response(code=500, msg=f"Failed to register with Bisheng: {str(e)}")
    # try:
    #     bisheng_info = await bisheng_service.register(user.username, user.password)
    # except Exception as e:
    #     return Response(code=500, msg=f"Failed to register with Bisheng: {str(e)}")

    # 注册到ragflow
    try:
@@ -47,7 +47,7 @@

    # 存储用户信息
    hashed_password = pwd_context.hash(user.password)
    db_user = UserModel(username=user.username, hashed_password=hashed_password, email=ragflow_info.get("email",  f"{user.username}@example.com"),ragflow_id=ragflow_info.get("id"),bisheng_id=bisheng_info.get("user_id"))
    db_user = UserModel(username=user.username, hashed_password=hashed_password, email=ragflow_info.get("email",  f"{user.username}@example.com"),ragflow_id=ragflow_info.get("id"),bisheng_id="")
    db_user.password = db_user.encrypted_password(user.password)
    db.add(db_user)
    db.commit()
@@ -61,14 +61,14 @@
    if not user:
        return Response(code=400, msg="Incorrect username or password")

    bisheng_service = BishengService(settings.sgb_base_url)
    # bisheng_service = BishengService(settings.sgb_base_url)
    ragflow_service = RagflowService(settings.fwr_base_url)

    # 登录到毕昇
    try:
        bisheng_token = await bisheng_service.login(login_data.username, login_data.password)
    except Exception as e:
        return Response(code=500, msg=f"Failed to login with Bisheng: {str(e)}")
    # try:
    #     bisheng_token = await bisheng_service.login(login_data.username, login_data.password)
    # except Exception as e:
    #     return Response(code=500, msg=f"Failed to login with Bisheng: {str(e)}")

    # 登录到ragflow
    try:
@@ -79,7 +79,7 @@
    # 创建本地token
    access_token = create_access_token(data={"sub": user.username, "user_id": user.id})

    upsert_token(db, user.id, access_token, bisheng_token, ragflow_token)
    upsert_token(db, user.id, access_token, "", ragflow_token)

    return Response(code=200, msg="Login successful", data={
        "access_token": access_token,

 app/api/chat.py

@@ -6,11 +6,13 @@
import asyncio
import websockets
from sqlalchemy.orm import Session
from starlette.responses import PlainTextResponse

from Log import logger
from app.api import get_current_user_websocket
from app.config.config import settings
from app.config.const import IMAGE_TO_TEXT, DOCUMENT_TO_REPORT, DOCUMENT_TO_CLEANING
from app.config.const import IMAGE_TO_TEXT, DOCUMENT_TO_REPORT, DOCUMENT_TO_CLEANING, DOCUMENT_TO_REPORT_TITLE, \
    DOCUMENT_TO_TITLE, DOCUMENT_IA_QUESTIONS
from app.models.agent_model import AgentModel, AgentType
from app.models.base_model import get_db
from app.models.user_model import UserModel
@@ -37,11 +39,14 @@
    print(f"Client {agent_id} connected")

    agent = db.query(AgentModel).filter(AgentModel.id == agent_id).first()
    print(agent_id)
    if not agent:
        ret = {"message": "Agent not found", "type": "close"}
        await websocket.send_json(ret)
        return

    agent_type = agent.agent_type
    print(agent_type)
    if chat_id == "" or chat_id == "0":
        ret = {"message": "Chat ID not found", "type": "close"}
        await websocket.send_json(ret)
@@ -339,10 +344,17 @@
                        except Exception as e:
                            logger.error(e)
                        # complete_response = ""

                        files = []
                        if upload_file_id:
                            files.append({
                                "type": "image",
                                "transfer_method": "local_file",
                                "url": "",
                                "upload_file_id": upload_file_id
                            })
                        answer_str = ""
                        async for rag_response in dify_service.chat(token, current_user.id, question, upload_file_id,
                                                                    conversation_id):
                        async for rag_response in dify_service.chat(token, current_user.id, question, files,
                                                                    conversation_id, {}):
                            # print(rag_response)
                            try:
                                if rag_response[:5] == "data:":
@@ -419,14 +431,20 @@
                        print(f"Received from client {chat_id}: {receive_message}")
                        upload_files = receive_message.get('upload_files', [])
                        title = receive_message.get('title', "")
                        sub_titles = receive_message.get('sub_titles', "")
                        workflow_type = receive_message.get('workflow', 1)
                        if not upload_files:
                            await websocket.send_json({"message": "Invalid request", "type": "error"})
                            continue
                        title_number = receive_message.get('title_number', 8)
                        title_style = receive_message.get('title_style', "")
                        title_query = receive_message.get('title_query', "")
                        if upload_files:
                            title_query = "start"
                        # if not upload_files:
                            # await websocket.send_json({"message": "Invalid request", "type": "error"})
                            # continue
                        try:
                            session = SessionService(db).create_session(
                                chat_id,
                                title,
                                title if title else title_query,
                                agent_id,
                                AgentType.DIFY,
                                current_user.id
@@ -446,15 +464,227 @@
                            })
                        if workflow_type == 1:
                            inputs["input_files"] = files
                        if workflow_type == 2:
                        elif workflow_type == 2:
                            inputs["file_list"] = files
                            inputs["Completion_of_main_indicators"] = title
                            token = DfTokenDao(db).get_token_by_id(DOCUMENT_TO_REPORT)
                            inputs["sub_titles"] = sub_titles
                            token = DfTokenDao(db).get_token_by_id(DOCUMENT_TO_REPORT_TITLE)
                            if not token:
                                await websocket.send_json(
                                    {"message": "Invalid token document_to_cleaning", "type": "error"})
                                    {"message": "Invalid token document_to_report", "type": "error"})
                        elif workflow_type == 3:
                            inputs["file_list"] = files
                            inputs["number_of_title"] = title_number
                            inputs["title_style"] = title_style
                            token = DfTokenDao(db).get_token_by_id(DOCUMENT_TO_TITLE)
                            if not token:
                                await websocket.send_json(
                                    {"message": "Invalid token document_to_title", "type": "error"})

                        complete_response = ""
                        async for rag_response in dify_service.workflow(token, current_user.id, inputs):
                        if workflow_type == 1 or workflow_type == 2:
                            async for rag_response in dify_service.workflow(token, current_user.id, inputs):
                                # print(rag_response)
                                try:
                                    if rag_response[:5] == "data:":
                                        # 如果是，则截取掉前5个字符，并去除首尾空白符
                                        complete_response = rag_response[5:].strip()
                                    elif "event: ping" in rag_response:
                                        continue
                                    else:
                                        # 否则，保持原样
                                        complete_response += rag_response
                                    try:
                                        data = json.loads(complete_response)
                                        complete_response = ""
                                        if data.get("event") == "node_started" or data.get("event") == "node_finished":  # "event": "message_end"
                                            if "data" not in data or not data["data"]:  # 信息过滤
                                                logger.error("非法数据--------------------")
                                                logger.error(data)
                                                continue
                                            else:  # 正常输出
                                                answer = data.get("data", "")
                                                if isinstance(answer, str):
                                                    logger.error("----------------未知数据--------------------")
                                                    logger.error(data)
                                                    continue
                                                elif isinstance(answer, dict):

                                                    message = answer.get("title", "")

                                                result = {"message": message, "type": "system"}
                                        elif data.get("event") == "workflow_finished":
                                            answer = data.get("data", "")
                                            if isinstance(answer, str):
                                                logger.error("----------------未知数据--------------------")
                                                logger.error(data)
                                                result = {"message": "", "type": "close", "download_url": ""}
                                            elif isinstance(answer, dict):
                                                download_url = ""
                                                outputs = answer.get("outputs", {})
                                                if outputs:
                                                    message = outputs.get("output", "")
                                                    download_url = outputs.get("download_url", "")
                                                else:
                                                    message = answer.get("error", "")

                                                result = {"message": message, "type": "message", "download_url": download_url}
                                                try:
                                                    SessionService(db).update_session(chat_id,
                                                                                      message={"role": "assistant",
                                                                                               "content": {
                                                                                                   "answer": message,
                                                                                                   "download_url": download_url}},
                                                                                      conversation_id=data.get(
                                                                                          "conversation_id"))
                                                except Exception as e:
                                                    logger.error("保存dify的会话异常！")
                                                    logger.error(e)
                                                try:
                                                    await websocket.send_json(result)
                                                except Exception as e:
                                                    logger.error(e)
                                                    logger.error("返回客户端消息异常!")
                                                result = {"message": "", "type": "close", "download_url": ""}


                                        else:
                                            continue
                                        try:
                                            await websocket.send_json(result)
                                        except Exception  as e:
                                            logger.error(e)
                                            logger.error("返回客户端消息异常!")
                                        complete_response = ""
                                    except json.JSONDecodeError as e:
                                        print(f"Error decoding JSON: {e}")
                                        # print(f"Response text: {text}")
                                except Exception as e2:
                                    result = {"message": f"内部错误： {e2}", "type": "close"}
                                    await websocket.send_json(result)
                                    print(f"Error process message of ragflow: {e2}")
                        elif workflow_type == 3:
                            image_list = []
                            # print(inputs)
                            complete_response = ""
                            async for rag_response in dify_service.chat(token, current_user.id, title_query, [],
                                                                        conversation_id, inputs):
                                print(rag_response)
                                try:
                                    if rag_response[:5] == "data:":
                                        # 如果是，则截取掉前5个字符，并去除首尾空白符
                                        complete_response = rag_response[5:].strip()
                                    elif "event: ping" in rag_response:
                                        continue
                                    else:
                                        # 否则，保持原样
                                        complete_response += rag_response
                                    try:
                                        data = json.loads(complete_response)
                                        complete_response = ""
                                        if data.get("event") == "node_started" or data.get(
                                                "event") == "node_finished":  # "event": "message_end"
                                            if "data" not in data or not data["data"]:  # 信息过滤
                                                logger.error("非法数据--------------------")
                                                logger.error(data)
                                                continue
                                            else:  # 正常输出
                                                answer = data.get("data", "")
                                                if isinstance(answer, str):
                                                    logger.error("----------------未知数据--------------------")
                                                    logger.error(data)
                                                    continue
                                                elif isinstance(answer, dict):

                                                    message = answer.get("title", "")

                                                result = {"message": message, "type": "system"}
                                        elif data.get("event") == "message":
                                            message = data.get("answer", "")
                                            # try:
                                            #     msg_dict = json.loads(answer)
                                            #     message = msg_dict.get("output",  "")
                                            # except Exception as e:
                                            #     print(e)
                                            #     continue
                                            result = {"message": message, "type": "message",
                                                      "download_url": ""}
                                            try:
                                                SessionService(db).update_session(chat_id,
                                                                                  message={"role": "assistant",
                                                                                           "content": {
                                                                                               "answer": message,
                                                                                               "download_url": ""}},
                                                                                  conversation_id=data.get(
                                                                                      "conversation_id"))
                                            except Exception as e:
                                                logger.error("保存dify的会话异常！")
                                                logger.error(e)
                                            # try:
                                            #     await websocket.send_json(result)
                                            # except Exception as e:
                                            #     logger.error(e)
                                            #     logger.error("返回客户端消息异常!")

                                        elif data.get("event") == "message_end":
                                            result = {"message": "", "type": "close", "download_url": ""}
                                        else:
                                            continue
                                        try:
                                            await websocket.send_json(result)
                                        except Exception as e:
                                            logger.error(e)
                                            logger.error("dify返回客户端消息异常!")
                                        complete_response = ""
                                    except json.JSONDecodeError as e:
                                        print(f"Error decoding JSON: {e}")
                                        # print(f"Response text: {text}")
                                except Exception as e2:
                                    result = {"message": f"内部错误： {e2}", "type": "close"}
                                    await websocket.send_json(result)
                                    print(f"Error process message of ragflow: {e2}")
                elif agent.type == "documentIa":
                    print(122112)
                    token = DfTokenDao(db).get_token_by_id(DOCUMENT_IA_QUESTIONS)
                    # print(token)
                    if not token:
                        await websocket.send_json({"message": "Invalid token", "type": "error"})

                    while True:
                        conversation_id = ""
                        # print(4343)
                        receive_message = await websocket.receive_json()
                        print(f"Received from client {chat_id}: {receive_message}")
                        upload_file_id = receive_message.get('upload_file_id', [])
                        question = receive_message.get('message', "")
                        if not question and not image_url:
                            await websocket.send_json({"message": "Invalid request", "type": "error"})
                            continue
                        try:
                            session = SessionService(db).create_session(
                                chat_id,
                                question,
                                agent_id,
                                AgentType.DIFY,
                                current_user.id
                            )
                            conversation_id = session.conversation_id
                        except Exception as e:
                            logger.error(e)
                        # complete_response = ""
                        files = []
                        for fileId in upload_file_id:
                            files.append({
                                "type": "document",
                                "transfer_method": "local_file",
                                "url": "",
                                "upload_file_id": fileId
                            })

                        answer_str = ""
                        complete_response = ""
                        async for rag_response in dify_service.chat(token, current_user.id, question, files,
                                                                    conversation_id, {}):
                            # print(rag_response)
                            try:
                                if rag_response[:5] == "data:":
@@ -467,8 +697,8 @@
                                    complete_response += rag_response
                                try:
                                    data = json.loads(complete_response)
                                    complete_response = ""
                                    if data.get("event") == "node_started" or data.get("event") == "node_finished":  # "event": "message_end"
                                    if data.get("event") == "node_started" or data.get(
                                            "event") == "node_finished":  # "event": "message_end"
                                        if "data" not in data or not data["data"]:  # 信息过滤
                                            logger.error("非法数据--------------------")
                                            logger.error(data)
@@ -484,6 +714,11 @@
                                                message = answer.get("title", "")

                                            result = {"message": message, "type": "system"}
                                            continue
                                    elif data.get("event") == "message":  # "event": "message_end"
                                        # 正常输出
                                        answer = data.get("answer", "")
                                        result = {"message": answer, "type": "stream"}
                                    elif data.get("event") == "workflow_finished":
                                        answer = data.get("data", "")
                                        if isinstance(answer, str):
@@ -494,12 +729,13 @@
                                            download_url = ""
                                            outputs = answer.get("outputs", {})
                                            if outputs:
                                                message = outputs.get("output", "")
                                                download_url = outputs.get("download_url", "")
                                                message = outputs.get("answer", "")
                                                # download_url = outputs.get("download_url", "")
                                            else:
                                                message = answer.get("error", "")

                                            result = {"message": message, "type": "message", "download_url": download_url}
                                            # result = {"message": message, "type": "message",
                                            #           "download_url": download_url}
                                            try:
                                                SessionService(db).update_session(chat_id,
                                                                                  message={"role": "assistant",
@@ -511,15 +747,16 @@
                                            except Exception as e:
                                                logger.error("保存dify的会话异常！")
                                                logger.error(e)
                                            await websocket.send_json(result)
                                            result = {"message": "", "type": "close", "download_url": ""}

                                            # await websocket.send_json(result)
                                        continue
                                    elif data.get("event") == "message_end":
                                        result = {"message": "", "type": "close"}

                                    else:
                                        continue
                                    try:
                                        await websocket.send_json(result)
                                    except Exception  as e:
                                    except Exception as e:
                                        logger.error(e)
                                        logger.error("返回客户端消息异常!")
                                    complete_response = ""

 app/api/files.py

@@ -10,6 +10,7 @@

from app.api import Response, get_current_user, ResponseList
from app.config.config import settings
from app.config.const import DOCUMENT_TO_REPORT_TITLE, IMAGE_TO_TEXT
from app.models.agent_model import AgentType, AgentModel
from app.models.base_model import get_db
from app.models.user_model import UserModel
@@ -99,7 +100,7 @@
    elif agent.agent_type == AgentType.DIFY:
        dify_service = DifyService(base_url=settings.dify_base_url)
        if agent.type == "imageTalk":
            token = DfTokenDao(db).get_token_by_id("image_and_text_conversion")
            token = DfTokenDao(db).get_token_by_id(IMAGE_TO_TEXT)
            if not token:
                raise HTTPException(status_code=500, detail="获取token失败，image_and_text_conversion！")
            file = file[0]
@@ -112,8 +113,8 @@
                data = await dify_service.upload(token, file.filename, file_content, current_user.id)
            except Exception as e:
                raise HTTPException(status_code=500, detail=str(e))
        elif agent.type == "reportWorkflow":
            token = DfTokenDao(db).get_token_by_id("document_to_report")
        elif agent.type == "reportWorkflow" or agent.type == "documentIa":
            token = DfTokenDao(db).get_token_by_id(DOCUMENT_TO_REPORT_TITLE)
            if not token:
                raise HTTPException(status_code=500, detail="获取token失败，document_to_report！")
            result = []

 app/api/public_api.py

@@ -5,7 +5,8 @@
from Log import logger
from app.api import Response
from app.api.auth import login
from app.config.const import IMAGE_TO_TEXT, DOCUMENT_TO_CLEANING, DOCUMENT_TO_REPORT, DIFY, BISHENG, RAGFLOW
from app.config.const import IMAGE_TO_TEXT, DOCUMENT_TO_CLEANING, DOCUMENT_TO_REPORT, DIFY, BISHENG, RAGFLOW, \
    DOCUMENT_IA_QUESTIONS, DOCUMENT_TO_TITLE, DOCUMENT_TO_REPORT_TITLE
from app.models.base_model import get_db
from app.models.public_api_model import DfToken, AppRegister
from app.service.common.api_token import DfTokenDao
@@ -17,7 +18,7 @@

@public_api.post("/sync/df_token", response_model=Response)
async def sync_df_token(df: DfToken, db=Depends(get_db)):
    token_dict = {IMAGE_TO_TEXT: df.image, DOCUMENT_TO_CLEANING: df.clean, DOCUMENT_TO_REPORT: df.report}
    token_dict = {IMAGE_TO_TEXT: df.image, DOCUMENT_TO_CLEANING: df.clean, DOCUMENT_TO_REPORT: df.report, DOCUMENT_TO_REPORT_TITLE: df.reoprt_title, DOCUMENT_TO_TITLE: df.title, DOCUMENT_IA_QUESTIONS: df.question}

    try:
        for api_id, token in token_dict.items():

 app/config/config.py

@@ -31,7 +31,7 @@
        kwargs['fwr_base_url'] = kwargs.get('fwr_base_url', '').replace('127.0.0.1', host_ip)
        kwargs['sgb_db_url'] = kwargs.get('sgb_db_url', '').replace('127.0.0.1', host_ip)
        kwargs['fwr_db_url'] = kwargs.get('fwr_db_url', '').replace('127.0.0.1', host_ip)
        kwargs['dify_base_url'] = kwargs.get('dify_base_url', '').replace('127.0.0.1', host_ip)
        # kwargs['dify_base_url'] = kwargs.get('dify_base_url', '').replace('127.0.0.1', host_ip)
        kwargs['basic_base_url'] = kwargs.get('basic_base_url', '').replace('127.0.0.1', host_ip)
        # Check if all required fields are provided and set them
        for field in self.__annotations__.keys():

 app/config/config.yaml

@@ -1,8 +1,9 @@
secret_key: your-secret-key
secret_key: Dt05gJnaqpT3Lx1iyWPrEolCKeXMbIcOSm28
sgb_base_url: http://192.168.20.119:13001
sgb_websocket_url: ws://192.168.20.119:13001
fwr_base_url: http://192.168.20.119:11080
database_url: mysql+pymysql://root:infini_rag_flow@192.168.20.119:5455/199c90c51171
database_url: mysql+pymysql://root:rag_gateway@192.168.20.119:3308/rag_gateway
#database_url: mysql+pymysql://root:infini_rag_flow@192.168.20.119:5455/199c90c51171
sgb_db_url: mysql+pymysql://root:1234@192.168.20.119:13306/bisheng
fwr_db_url: mysql+pymysql://root:infini_rag_flow@192.168.20.119:15455/rag_flow
PUBLIC_KEY: |

 app/config/const.py

@@ -3,7 +3,9 @@
DOCUMENT_TO_CLEANING = "document_to_cleaning"
DOCUMENT_TO_REPORT = "document_to_report"
IMAGE_TO_TEXT = "image_and_text_conversion"

DOCUMENT_TO_REPORT_TITLE = "document_to_report_title"
DOCUMENT_TO_TITLE = "document_to_title"
DOCUMENT_IA_QUESTIONS = "document_ia_questions"

### -----------app register --------------
RAGFLOW = "ragflow_app"

 app/models/public_api_model.py

@@ -53,6 +53,9 @@
    image: Optional[str] = ""
    clean: Optional[str] = ""
    report: Optional[str] = ""
    reoprt_title: Optional[str] = ""
    title: Optional[str] = ""
    question: Optional[str] = ""


class AppRegister(BaseModel):

 app/models/token_model.py

@@ -23,8 +23,8 @@
    # 参数验证
    if not isinstance(user_id, int) or user_id <= 0:
        return
    if not access_token or not bisheng_token or not ragflow_token:
        return
    # if not access_token or not bisheng_token or not ragflow_token:
    #     return
    db_token = None
    try:
        # 查询现有记录

 app/service/difyService.py

@@ -65,28 +65,22 @@
                raise Exception("Authorization header not found in response")
            return authorization

    async def chat(self, token: str, user_id: int,  message: str, upload_file_id: str, conversation_id: str):
    async def chat(self, token: str, user_id: int,  message: str, files: list, conversation_id: str, inputs:dict):

        target_url = f"{self.base_url}/v1/chat-messages"
        files = []
        if upload_file_id:
            files = [
                {
                    "type": "image",
                    "transfer_method": "local_file",
                    "url": "",
                    "upload_file_id": upload_file_id
                }
            ]
        data = {
            "inputs": {},
            "inputs": inputs,
            "query": message,
            "response_mode": "streaming",
            "conversation_id": conversation_id,
            "user": str(user_id),
            "files": files
        }

        # if parent_message_id:
        # data["parent_message_id"] = parent_message_id
        # print("----------------------------------------------")
        # print(data)
        # print("-----------------------------------------------")
        async with httpx.AsyncClient(timeout=300.0) as client:
            headers = {
                'Content-Type': 'application/json',

 app/task/fetch_agent.py

@@ -113,12 +113,12 @@
            db.commit()  # 提交事务

        initial_agents = [
            ('80ee430a-e396-48c4-a12c-7c7cdf5eda51', 1, '报告生成', 'BISHENG', 'report'),
            # ('80ee430a-e396-48c4-a12c-7c7cdf5eda51', 1, '报告生成', 'BISHENG', 'report'),
            ('basic_excel_merge', 2, '报表合并', 'BASIC', 'excelMerge'),
            ('bfd090d589d811efb3630242ac190006', 4, '文档智能', 'BISHENG', 'report'),
            ('7638f006-38a2-4c21-a68e-c6c49b304a35', 4, '文档智能', 'DIFY', 'documentIa'),
            ('da3451da89d911efb9490242ac190006', 3, '知识问答', 'RAGFLOW', 'knowledgeQA'),
            ('e96eb7a589db11ef87d20242ac190006', 5, '智能问答', 'RAGFLOW', 'chat'),
            ('basic_report_clean', 6, '文档报告', 'DIFY', 'reportWorkflow')
            ('basic_report_clean', 6, '报告生成', 'DIFY', 'reportWorkflow'),
            # ('basic_excel_talk', 6, '智能数据', 'BASIC', 'excelTalk'),
            # ('basic_question_talk', 7, '出题组卷', 'BASIC', 'questionTalk'),
            # ('9d75142a-66eb-4e23-b7d4-03efe4584915', 8, '小数绘图', 'DIFY', 'imageTalk'),
@@ -140,10 +140,10 @@

def sync_agents():
    try:
        bisheng_data = get_data_from_bisheng(BISHENG_NAMES_TO_SYNC)
        # bisheng_data = get_data_from_bisheng(BISHENG_NAMES_TO_SYNC)
        ragflow_data = get_data_from_ragflow(RAGFLOW_NAMES_TO_SYNC)

        update_ids_in_local(bisheng_data)
        # update_ids_in_local(bisheng_data)
        update_ids_in_local(ragflow_data)

        print("Agents synchronized successfully")